Discriminative training methods for hidden markov models: Theory and experiments with perceptron algorithms Export

@inproceedings{Collins:2002, abstract = {We describe new algorithms for training tagging models, as an alternative to maximum-entropy models or conditional random fields (CRFs). The algorithms rely on Viterbi decoding of training examples, combined with simple additive updates. We describe theory justifying the algorithms through a modification of the proof of convergence of the perceptron algorithm for classification problems. We give experimental results on part-of-speech tagging and base noun phrase chunking, in both cases showing improvements over results for a maximum entropy tagger.}, author = {Collins, Michael}, booktitle = {EMNLP}, citeulike-article-id = {1459587}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.18.6725}, keywords = {perceptron-learning-rule, phd}, posted-at = {2009-11-11 13:52:30}, priority = {0}, title = {Discriminative training methods for hidden markov models: Theory and experiments with perceptron algorithms}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.18.6725}, year = {2002} }

TY - CONF ID - Collins:2002 L3 - citeulike-article-id:1459587 N2 - We describe new algorithms for training tagging models, as an alternative to maximum-entropy models or conditional random fields (CRFs). The algorithms rely on Viterbi decoding of training examples, combined with simple additive updates. We describe theory justifying the algorithms through a modification of the proof of convergence of the perceptron algorithm for classification problems. We give experimental results on part-of-speech tagging and base noun phrase chunking, in both cases showing improvements over results for a maximum entropy tagger. TI - Discriminative training methods for hidden markov models: Theory and experiments with perceptron algorithms T2 - EMNLP KW - perceptron-learning-rule KW - phd AU - Collins, Michael PY - 2002/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.18.6725 ER -